Measuring and Modelling the Reliability of the Transport Supply Experiments with Swiss Respondents

Abstract

Current transport models focus mainly on the variables travel time and travel costs to explain choices made by travellers. In contrast to this, research results substantiate the influence of a range of additional explanatory variables. The reliability of the transportation system is one important example. The inclusion of those variables into the forecasting of the traveller decision requires more knowledge about their valuations. In this context this study pursues various aims: First, to compare the influence of the reliability of travel times with variables normally included; to confirm the plausibility of the available empirical results; to describe the impact of variables representing the reliability; to recommend an appropriate implementation into discrete choice models; finally reliability should be valued monetarily and a recommendation has to be given for its inclusion the cost-benefit-analysis. This study gives an overview of the state of the art and recent empirical work the valuation of travel time and reliability. Following the design, the structure and the performance of two stated preference (SP) surveys are explained. Within these case studies different approaches of presenting reliability are applied. The used SP-experiments employ route choice, mode choice and departure time choice stated preference. This first part concludes with a descriptive analysis of the sample. The second part contexts develops discrete choice models for each of the choice experiments. The estimations start with basic models which will be expanded and finish with a recommended model for each type of experiment. The choice analysis allows the estimation of a value of travel time savings and a value of reliability. The results of the study show that travellers value the increase travel time reliability and the saving of travel time equally. This substantiates the importance of considering both those variables in choice models. The reliability can be implemented in each choice context, route choice, mode choice and departure time choice. The highest goodness of fit is observed for the route choice models. It is recommended that the reliability of travel times on routes should be defined as the maximum delay and also as the probability of this duration. The linear formulation of these variables in the utility function leads to the highest precision. In contrast to experiences earlier studies models with mean travel time or mean delay and their variance do not raise the model fitness. Further it is recommended that the willingness to pay for reliable travel times should be calculated by a model formulation which includes reliability with an interaction form between the maximum delay and the elasticity of the lateness probability. The estimation process brings up an equation that can be used in cost-benefit-analysis to calculate a monetary value for different reliabilities by a change of supply. The willingness to pay for a 50% reduction of the lateness probability per minute of a delay is estimated 0.28 CHF for car drivers and 0.12 CHF for PT users. The willingness to pay for a complete cut of lateness is about 30% higher than the estimated value of travel time savings. The Estimation shows no significant difference of the reliability valuation between market segments like trip purposes or distances. The chosen approach does not allow calculating the willingness to pay for the reduction of earlier arrivals For this topic further work has to be done.